Volume 300. number 2.

175-178 FERS 10865 March 1992

Q 1992 Federation or European Biochemical Societies W145793/92/55,W

A theoretical method to predict DNA permutation gel electrczphoresis

from the sequence

D. Boffelli”, P. De Santis’, A. Palfeschi”, G. Risuleob and M. Savinob

“Diparlimcnlo di Ch?nica. Lhiversil~ di Romu ‘L-GSapienza’, P. le A. More 5, 00185 Rorntz,Italy and bDiparrimenfo di Genetica
c BioIogia Molecolare, Universiki di RO~Q ‘La Sapienza’ P. Ie A. More S, 00185 &ma, &al_1

Received 15 January 1992

The gel clectrophoretic permutation assays of DNA fiagmcntscxperimentally investigated by different authors were theoretically reproduced using
our theoreti~l modei oF~~uen~d~cndent curvature. The general pattern ofagreement obtained suggeststhat our methodcan be usefully adopted
as an alternative IO the cx~~menta1 assay. in pa~~~ular where the Jack of a suB%5ent number of unique restriction sites in the Fragment prevents
the correct localization of the main bend site.

Thcorctical prediction: DNA curvature: DNA permutation: Gel electrophoresis

1. INTRODUCTlON fail where two or more bend sites in contrast of phase

are present in the DNA tract.
After the detection of sequenrf -‘-pendent cur~%ttre A few years ago we theoretically investigated the
in K-DNA, a kinetoplast DNA fragment of L. IU- physical basis of the sequen~dependent curvature in
renrohe [I], many DNA tracts were shown to bc in- DNA and found that the nearest-neighbour di~erential
trinsically curved, as indicated by their atlo~lousty i~ltera~tion between base pairs could suffice to explain
slow electrophoretic mobility through polyacrylamide the origin of su~rstructures in DNAs [14-19], as ori-
gels El-121. ginally suggested by Trifonov (20.21J.
This supers~ructural aspect of the double hciix has, in Among these investigations, we found electrophoretic
fact, been found in biologically relevant DNA tracts retardation to be linearly correlated in first approxima-
~rnpl~cat~ in regulation, recombination and replication tion with the dispersion of curvature (cr’) along the
processes, and seems to be involved in the phasing of sequence:. This is proved by the good agreement
nu~leo~omes [ 131. (correlation factor = 0.98) between a’ and the corrc-
While it appears to be relatively easy to detect the sponding electrophoretic retardation (ranging in value
presence of curvature in DNA fragments from their between t .O and 3.3), of about 450 multimcric synthetic
electrophorctic anomalies, it is, however, more difficult DNA oligomers which differ in Fcriodicity. sequence
to locate bend sites. Scvcral years ago Wu and Crothers and length (So-200 bp) [i&-19].
[3] devised an ingenious electrophoretic assay to map This suc^css prviilptcd us to thcorc. ~11~reproduce
the bending site (where a unique m, ;n curvature is yrcs- the permutation assays of DNA tracts experimcntaily
cnt) in a DNA fragment. The technique involves digcst- investigated. even if their fengths might ovcrcomc the
ing tandem dimer fragments at unique restriction en- molccufar complexity of the synthetic DNAs inves-
donuclcase sites and plotting the gel elcctrophorc~’ tigated and the clcctrophorctic conditions used (per-
retardation of the fragmeilts against their base-p:lli centage of acrylamidc and tcmpcraturc) were not stand-
position. The possible bend sit2 should correspond to ardized. In fact it must be considered that in pcmuta-
the intcrpolatcd minimum of this rctardkttion plot. tion gel clcctrophorcsis only the trend of retardation
Curvature sites of scvcral DNA tracts have been against the restriction silts along the scqucnce is rc-
identified using this technique, and permutation gel yuircd for locating the main bend silt.
clcctrophorcsis has bccomc a powerful tool to invcsti-
gate the implications of curvature in biologically rc-
lcvant DNA fragments when a suitable set of single
restriction sites is prcscnt. The tcchniquc, howcvcr. can

I75
 angular deviations of the local helical axis (practically
the perpendicular to each base pair plane) from t!k
average diion.
This corresponds in the complex plane representation
to:

$= < an) 1 C(n) > - c c(n) > < CIn) >

where:

is the integml curvature between the first and the n-th

nucleotide step In is the total number of bp in the con-
sidered tract).
We found d to be linearly correlated with gel elec-
trophoretic retardation, quantified as the ratio between
apparent and real lengths. This allows the easy calcula-
tion of an analog of the permutation assay by plotting
0” for the cyclic permuted sequences.

3. RESULTS AND DISCUSSION

We have adopted our theoretical model to predict the
experimental permutation assays of several DNA frag-
ments (180-750 bp in length) investigated by different
authors. The experimental points were obtained from
the corresponding illustrations of the original papers by
Ot---.‘--.*‘----‘--*.f----f~
m mo am Io(R awJD visual interpolation.
0
A general pattern of agreement is obtained, in parti-
Fig I. -1 pattern of the pcrlnutatiott zieisil>% of SW0 fta~cnts cular as the common trends of ti and retardation
imtw+ati by Milton and Geteland 1111compared with the cwva.
wfm diagramof the oh& sequntcecatcularedover lccunrntIran5
profiles are concerned, whereas the absolute values are
of IOObp. as ~rtl as &th the pattcm of thcmcticai assays. not always reproduced because sf the lack of stand-
ardized experimental conditions (percentage of acry-
lamidc and temperature), as well as of the high mo!Lzu-
tar weight which plausibly amplifies the approximation
defects of the model. Nevertheless. the positions of the
curved regions along the sequence were correctly
where: g I 5 rhc inrcgmion grid, I@ = lU.4 is the perio- predicted in all the cases.
dicity of B-DNA, and 4s) = Ip-ir’ is the deviation of Fig. 1 show the five tr::cts corresponding to s3me
the dinucleotide step at sequence number s from the anomafous fragments of restricted SW0 investigated by
canonical II-DNA, as theoretically evaluated by con- Milton and Gesteland [l 13. The trcrtds 4~:SxpcrimcntJ
focmational energy cakulations. and tkoretical permutation assays ate repoAle4 along
Curvature was evaluated over 3 recurrent turns of the sequence. In the same figure the positions of the bent
EDNA (g = 31 bp) to reduce the noise and ported regions are represented in the curvature diagram,
s.%tsusthe !x.qmce number n as ‘curvature function’, averaged over rccurmnt tracts of 100 bp. It is easy to
Thinsdiagnm allows an easy visualization of curvature obscrvc that the minima of permutation assays cor-
sites along DNA comzsponding to the maxima of the respond to the maxima of the curvlriure diagram which
cuI%?ture protile, also shows the pres~tr;i: of some regions of SV40 DNA
Starting from this repnzentation of the superstrvc- not considered by the cited authors.
tuwl fcaturrr 0fDl4A it is czq to rcproducc a suitable Fig 2 illustrate the case of the curved DNA at the
thre+dimenGonal rcp~~~tation of the DNA super- upstream activation sites (UASI of &-ccll-spccific genti
strueturt as a writhing of the helix axis. assigning the in yeast. investigated by Innkutrhi m a!. !!4 u&g dr-
rwduiw of curvature as a Ming angje and the pi9~~ cular permutation analysis of the 513 bp fragment.
t * a torsional the -pond- which contains IWO UAS r&ons around the main
curWture OilC.Thr two diagmms or the righht panrl
~RCYW the direct compz&on ktwccn the relative rn+
&lily and the cknp of the dispcrsi0n of cufvi4tufc
Volume 300. number 2 FEBS LE-i-fERS March 1992

PdTmts wa
0 1w 4w

Fig. 2. activation site in yeast and the corresponding stereoprojeetions of

e%~~rnent~~ [It] and thewetieal ~~~tati~n assays.

resulting from the cyclic parm~:~~on of the sequtince. riophage 178 bp fragment investigated by Zahn and
The trend of the two diagrams is practically the same Blatter [5]. In spite of the lack of restriction sites over
and the position of the minimum corresponds to the more than two thirds of the sequence, the authors local-
center of the curved tract as represented in the ized the maximum of the curvature at about n = 51.
stereoprojection as well as to the main maximum of the Such a position in fact should be shifted by one turn of
curvature diagram, both reported at the left pane1 of the a double helix, as clearly shown by the theoretical
figure. diagrams.
Fig. 3 shows the case of the ori region of R bactc- Finally, Fig. 4 illustrates the case an clectrophoreti-

1.60

l.BLI

1.80

1.10

a.10

f.Ol

a.9

LO

au

\ a.*

I 0.c

0.6
fig 4.(L& gartett Cunatulr dkqram of a 537 bp fragment orthc Polyoma virus and the concspondingstt.%opjection c?ithe writhing of the
beiii axis. (Right pan& Comparison bct~en experimental 1221and theolrtical permutation assays.

salty u~~~ul~~ tract cf Polyoma virus (I 761-2298 [4] Eussi. L. and Smith. D.M. (1984) Cell 39.643-652.
tract) investigated in Qur laboratary 1221. The frameat [q Ztthn. K. and Blattncr. F.R. (1985) Nature 317.451453.
f6] Kocpscl, R.R. and Khan. S.A. (1986) Science 233, 1316-1319.
was restricted with AceI, PruII, S##. Avali and Taqi
17 Deb. 5.. DeLucia. A.L.. Koff. A.. Tsui. S. and Tcgtmaycr. P.
afkr dime&km and cloning in pUCI8. Also in this (1986) Mol. Cell Biol. 6.4578-4584.
casz, irkspite of the rather unitbrm curvzture as shown [S] Snydc. M.. Buchman. A.R. and Davis, R.W. (1986) Nature 324.
by the stereoprojection. the retardation gel permutation 87-84.
diagram is t~~ti~lly ague. [9] Goursc. R.L.. dt Boer. H.A. and Homura. M. t 1986) Cell 44.
197 -205.
Apart from the casts illustrated in this paper. we McAllister. C.F. and Achbcrgcr. EC. t 1988)J. Biol. Chcm. 263.
analyzed at1 the ~~~tati~n assays app~~~n~ in the %t743-~1749.
litcraturc and WCgcnmlly found satisfactory results. hluton. D.L. and Gcsrcland. R.F. (19X8)Yuclcic Acids Rcs. 16.
tfQ~‘cvCr.a deeper analysis of cxperimcntal data shows 3931 3949.
lnokuchi. K.. Nakeyama. A. and Hishinuma. F. (19118) Nucleic
that in most c;jn the limited number of uniqur restric-
Acids Rcs. 16.6693 671 I,
tion sites bars the possibility of precisely I~cating curva- Boffctli. D., De Santis. P.. PaWxcir. A. and Savir,, . M. f1991)
ture sites. In some cases the interpolation of cxprimcn- Biophys. Chcm. 39. l27- 136.
tai data losates the hypothetical curvature mtn: on an IIr Samis. P.‘..Momti. S.. Patfeschi. A. and Savino. M. (1986)
incorrect DNA region with a conscqucnt mhlndcrstart- in: Structures and @namiis or Nucicic AC+.‘,. !‘rrltcin> nci.1
Membranes (E. Ctcmnti and S. Chin dls : pp. 3 L *i Phum
ding of the sequences responsible for DNA curvature.
Press.Mx York.
As a gcncml oondusion. the main result of this paper Dc Santil;, P.. Moroxtti. S.. Pallcschi. A.. Savino. M. and
is that Qur thconrtical method is a valid alternative to Sfipioni. A. 1lSlgS) in: Bj~i~i~l and Aniticial ~ntelliE~ee Sys-
permutation gel clectrophomsis in Order to kicatc the tcms tE. Clcmcnt~ and S. CXn cda.) pp. IQ”-IhI. ESCOM
curwd eons of DNA f@mznts. Scz~cc. NW York.
gl6] De Santir. P.. Pall~hi, A.. Saviw. M. and Scipioni, A. (1988)
Biophp C-hen. 32. MS-- 3f7.
al?j Dc Santix P.. Gallo, G.. P&exhi. A.. Savino. M. and Scipioni.
A. Ll’?WJ)J. Itloj. Liquid- 41. 291 394.
ltr;j Cacchionc. S.. De Samis. P.. E&r. D.. Pall~hr, A. and &inn.